Impact of Data Assimilation Variations on 1-km Forecasts of 24 May 2011 Tornadic Supercells

A tornado outbreak struck central and eastern Oklahoma on 24 May 2011. During the early part of this severe weather event, supercells were isolated, evolving independently, but later in the afternoon, storm and supercell coverage increased along with storm splits and mergers. The Center for Analysis and Prediction of Storms’ (CAPS) Advanced Regional Prediction System’s (ARPS) real-time 1-km forecasts had good success in forecasting the potential severity of this event, but the forecasts of storm-scale details might be improved through the use of different data assimilation (DA) techniques, especially in the later stages of the event. In this work, four variations of the incremental analysis updating (IAU) data assimilation scheme, including the original IAU applied equally to all variables, variable-dependent IAU, analysis cycling with the original IAU, and analysis cycling with the variable-dependent IAU, are examined and compared to equivalent runs with no IAU or analysis cycling steps. All of the IAU DA processes utilize increments produced from the 3DVAR and complex cloud analyses. For the original IAU, fractional increments are added during 5-min IAU windows using a temporally-weighted distribution that peaks in the middle of the IAU window. For the variable-dependent IAU, the temporally-weighted distributions of the fractional increments are varied for each variable. The simulations using the analysis cycling have two 5-min IAU windows. For all experiments, 2-h forecasts are produced subsequent to the DA steps for eight initialization times using ARPS with 1-km horizontal grid spacing. Observation-point, neighborhood, and object-based verification techniques are employed to assess the performance of the convective forecasts. For the object-based technique, six tornadoes from three storms of interest are used to verify simulated low-level circulations using the updraft helicity field. Results indicate that both analysis cycling and variable-dependent IAU can improve forecasts of near-surface variables and low-level circulation tracks.